Meta Kills Instagram AI Feature That Let Anyone Turn Public Accounts Into Deepfake Fodder

Meta has shut down an Instagram feature that let users generate AI images by simply tagging any public account, just three days after announcing it, following a swift and pointed backlash from safety advocates and Hollywood's actors union.

The feature was part of the rollout of Meta AI's new Muse Image model, announced Tuesday. The pitch was simple: type an @-mention of a public Instagram account into Meta AI, and the tool would pull from that account's content to generate images referencing it. Meta framed it as a creative tool. The problem was that it worked on any public account, whether that account's owner had agreed to it or not, and the only way out was to opt out manually through settings, after the fact.

That design, some critics said, put the burden on individuals to protect themselves from having their own likeness used in AI-generated content they never consented to.

Backlash arrived fast

Haley McNamara, executive director and chief strategy officer of the National Center on Sexual Exploitation, was blunt about the risk. "Not only does this obviously erode our rights to our own likeness… but it is an obvious tool for #sextortion and other scammers!" she said Friday, according to The Verge. She also criticized the underlying design choice directly: "Pursuing high-risk design & then putting the onus on individuals to jump through hoops to opt out is unacceptable."

The Screen Actors Guild moved even faster than Meta did. The union told its members to opt out of the feature and published instructions on how to do it, a notable move given how central likeness and consent issues have become to SAG's fights with tech companies over AI in recent years.

By Friday, Meta reversed course. In an update to its blog post about Muse, the company acknowledged the feature "missed the mark" and confirmed it's "no longer available." Meta said its original intent was to build a useful creative tool while still giving people control over whether their public content could be referenced this way, but didn't dispute that the opt-out mechanism fell short of that goal.

Why the opt-out wasn't enough

The core issue wasn't that Meta offered no controls at all. It's that the controls were opt-out rather than opt-in, meaning every public Instagram account was fair game for AI generation by default, and the account holder had to go find the setting and turn it off to be excluded. For ordinary users, that's an inconvenience. For public figures, celebrities, and anyone with a large following, it's a much bigger problem: their images could be remixed into AI content the moment the feature went live, before they'd even heard it existed.

That's precisely the scenario that alarms groups like the National Center on Sexual Exploitation, which has spent years warning about AI-generated sexual content and deepfake-enabled extortion schemes. Giving anyone the ability to tag a public account and generate imagery referencing that person, without prior consent, hands potential bad actors a low-friction path to exactly the kind of abuse those groups track.

A pattern for AI feature rollouts

The turnaround here is notable mostly for its speed. Meta announced the tagging feature on Tuesday and killed it by Friday, a three-day span that suggests the company either underestimated the reaction or moved unusually quickly once the criticism landed. It's also a familiar shape for how AI features tend to launch and then get walked back across the industry: ship first, framed around creative utility, then scramble to add guardrails once critics point out how easily the same tool doubles as a harassment vector.

Meta hasn't said whether it plans to reintroduce a version of the feature with different consent defaults, such as requiring accounts to opt in before being tagged. The blog update simply confirms the feature is off. Muse Image itself, the underlying AI model, isn't going anywhere. It's the @-mention shortcut into other people's public content that's been pulled.

For now, the episode is a reminder of how quickly consent questions can sink an AI feature, even one framed as harmless fun. The Screen Actors Guild's rapid response, paired with pressure from anti-exploitation advocates, shows that scrutiny of exactly whose data and likeness get fed into generative AI tools has become a fast, organized reflex rather than a slow-burn controversy. Whatever Meta builds next along these lines will likely need consent baked in from the start, not bolted on as a settings toggle after the backlash arrives.